Abstract: In order to solve the practical application problems, including abnormal data insufficiency and unavailability which often happen in bearing fault diagnosis application, one-class bearing fault detection using improved real-valued negative clone selection (RNCS) algorithm based on higher order statistics (HOS) is presented. In this model, only normal sample data are needed for training purposes. RNCS is used to generate probabilistically a set of fault detectors that can detect any abnormalities (including faults and damages) in the behavior pattern of bearings. The clone mature operator and self-adaptive mutation operator are adopted in order to improve detection rate of antibodies and the convergence rate. Further, as the extracted HOS feature matrixes from original signal are too abundant to make further intelligent detection and diagnosis using HOS, feature matrixes are transformed to singular value spectrums which are used as features for overcoming this problem. The behavior of the classifier based on parameter selection and number of normal training samples is analyzed. Comparison of the performance of detection of RNCS with different detector’s numbers is experimented. The proposed approach are compared with other detection techniques. The results show the relative effectiveness of the proposed classifiers in detection of the bearing condition with some concluding remarks.

Key words: Fault detection, Higher order statistics, Negative clone selection, Self-adaptive mutation operator, Singular value decomposition